Driven by technological advancements, modern railway systems are increasingly characterized by higher transport capacity and improved sustainability in response to resource constraints and climate change. At present, wheel–rail contact, as the dominant technical approach for railway transportation, serves to guide the train and provide the necessary traction and braking forces. However, this contact is susceptible to defects such as wheel flats, which can cause impact vibrations and noise, and pose safety risks. To effectively detect wheel flats, axle box acceleration (ABA) measurements hold great potential, as it has been widely adopted to monitor track structures and interface deterioration. This study is conducted to develop a reliable 3D finite element (FE) model to simulate wheel flat-induced impacts, and to investigate the effects of the rolling speed and flat parameters on the impact force and ABA responses.

To achieve the research objectives, a 3D FE model with a flat was developed based on the lab test data from V-Track. The finite element analysis was conducted through an implicit–explicit sequential method, and the developed FE model was validated against experimental measurements. With the validated model, a parametric study was conducted to study the effects of wheel rolling speed, flat length, and flat development stage on the impact force and ABA responses, with the latter two parameters investigated by adjusting the discretized tread profiles in the flat region. For all simulated and measured signals, time–frequency features were obtained using Continuous Wavelet Transform and Synchrosqueezed Wavelet Transform. The resulting wavelet power spectra can effectively capture localized energy variations induced by the wheel flat.

The results demonstrate that flat-induced wheel–rail impact force increases non-linearly with wheel rolling speed, and the correlation between the flat length and the impact force is not consistently positive. Moreover, the wear development of the flat leads to a noticeable reduction in impact force at low speeds. The ABA responses are influenced by a combination of factors, including speed, wavelength, and flat-related wheel and track modes. The conclusions and recommendations presented at the end of the thesis highlight the need for a more comprehensive study that considers a broader range of flat geometric parameters to support the development of flat monitoring and the formulation of ABA-based wheel assessment rules.